It has been desired to realize a solid-state image sensing device having a wide dynamic range that may enable shooting of details without losing details of high-brightness information even in backlit shooting with illumination light of headlights of cars and ball game grounds or sunlight or bringing an object image in the low-brightness part too much darker.
Under the circumstances, technologies of expanding dynamic ranges with respect to solid-state image sensing device such as CCDs are disclosed in Patent Document 1 (Japanese Patent No. 2125710), Patent Document 2 (JP-A-03-117281), Patent Document 3 (JP-A-09-205589), Patent Document 4 (JP-A-2004-320119), Patent Document 5 (JP-A-2005-278135), Patent Document 6 (JP-A-2010-284213), etc.
Patent Document 1 shows an example of expanding a dynamic range of a CCD by providing two or more plural regions (cells) having different sensitivity characteristics within one pixel of the CCD or the like to provide knee characteristics that input/output characteristics change in a stepwise fashion.
The knee characteristics refer to a characteristic curve shown by a relationship of the output current to the amount of exposure takes smaller values in a high-input region than those in a low-input region, and are often referred to as a high-brightness signal compression technology.
As methods of changing the photosensitivity of the photosensitive region (cell), for example, changing the aperture ratio of the device, providing an optical filter (ND filter), changing impurity concentration, etc. are described.
According to Patent Document 1, the example is applicable to an image sensing device of XY address-type other than the CCD, however, there is no detailed description.
Patent Document 2 discloses an example of realizing higher dynamic range without losing details with highlight of a light valve or the like using adjacent pixels in a photosensitive pixel cell of a CCD or cells with different photosensitivity characteristics as a pair and adding signal charge of the respective cells within one pixel using it as signal charge of the pixel.
In this case, as means for changing photosensitivity, cells with different pixel areas are paired, for example.
Patent Document 3 similarly divides one pixel of a sensitivity pixel cell of a CCD into two different regions, and mixes signal charge of the regions with different sensitivity of the same pixel in a vertical register and vertically transfers it. Further, in the technology, a video signal is formed by sorting the signal charge with different sensitivity into two horizontal transfer gates using a sorting gate, and clipping the signal at the high-sensitivity side using an external signal processing circuit and adding it to the signal at the low-sensitivity side.
In this case, the characteristic graph of the video signal output to the amount of incident light is a line graph, and the gradient is steep at the high-sensitivity side (low-illuminance side) and the gradient is mild at the low-sensitivity side (high-illuminance side).
Patent Document 4 discloses an improvement method, in an image sensing device including high-sensitivity imaging cells and low-sensitivity imaging cells, for a problem that RAW image data volume (raw data) becomes larger with data of both cells.
Specifically, whether or not recording of image information of the high-brightness part is necessary is automatically determined by analyzing shot image information. If the determination is “YES”, the RAW image data of the high-brightness part is recorded together with the information of the low-brightness part. If the determination is “NO”, the information of the high-brightness part is not recorded, but only the RAW image data of the low-brightness part is recorded.
One pixel is formed by combining a main photosensitive pixel cell (with larger area and higher sensitivity: mainly using the center part of a microlens) and a sub-photosensitive pixel cell (with smaller area and lower sensitivity: provided at the edge side of the microlens).
Patent Document 5 discloses a CMOS image sensor including a column-parallel ADC having a comparator and an up/down counter. The CMOS image sensor can execute addition operation of pixel digital values over plural rows without additional circuits such as an adder and a line memory device.
However, compared to the pixel having an area as a total of all areas of target pixels in the case of the divided pixel addition, in the case of division, ineffective regions (dead spaces) that do not directly contribute to photosensitivity are produced in signal processing.
Accordingly, the areas of the divided individual cells are smaller than in the case of simple division into four, the number of saturated electrons decreases compared to the former case and the Shot noise relatively increases, and S/N of the divided individual cells is deteriorated.
Since the Shot noise is also added at each time of addition, S/N of the result of the divisional addition is deteriorated.
Further, addition processing of pixel signals is analog signal addition and the sensitivity is different with respect to each pixel, and thus, there are problems that the saturation values vary and the break point positions vary.
Furthermore, in the case of digital addition, it is necessary to provide a memory outside of the sensor.
That is, in the existing addition method of dividing one pixel cell into two or more plural pixel cells with different sensitivity or accumulation times and measuring the sensitivity as the amount of saturation charge Qs of the pixels, the amount of saturation charge Qs varies with respect to each pixel. Accordingly, the addition result varies with respect to each pixel for the same amount of light.
In other words, in the sensitivity curves (line graphs) with the amount of incident light as a horizontal axis and the amount of saturation charge Qs as a vertical axis, the break point positions (vertical axis) vary at the divided pixel cell addition points (horizontal axis).
Accordingly, Patent Document 6 suggests a method of realizing a wide dynamic range by applying a technology of regarding four pixels as one pixel and varying respective accumulation times of the four pixels. In this technology, four signals are added.
FIG. 1 is a diagram for explanation of a method of realizing a wide dynamic range by varying respective accumulation times of four pixels.
In the method, pixels formed by respectively dividing single colors of R, G, B into four are used.
Further, as shown by signs A to D in G (green) on the upper left of FIG. 1 as an example, a structure having four pixels with different photosensitivity or amounts of exposure is assumed. In FIG. 1, FD shows a floating diffusion part.
Furthermore, by summing the outputs of the four pixels, a wide dynamic range is realized.
According to the technology, divided pixel addition without variations in the number of output electrons of pixels may be realized, and the wide dynamic range in which the sensitivity is higher in the lower amount of incident light, the sensitivity is lower with the higher incident light, and the output is not saturated can be provided.